About the Datasets

Fire

  • Refer to the dataset in_cali for fire points restricted to California’s borders.
  • Time Frame: The HMS system has operated 24x7x365 since June 16th, 2003, when some of its first satellites have been put to use. Some of these satellite sources have been discontinued after some time, but Terra/MODIS is one of the first satellites used and is still in use.
  • Data Source: Satellites used include GOES, Terra/MODIS, Aqua/MODIS, NOAA, MetOp, and S-NPP/VIIRs. Even though most of the satellites only operated for a certain time, we can still use HMS as our main data source because Terra/MODIS has been present since the beginning. This can be used as a baseline, and we can draw data from other satellites to compare with it if needed.

Data Structure, Variables, Etc.

  • They record fire detection data several times a day (around once every 2 hours), but there can be a little delay in making the data available (this doesn’t affect the observation time recorded).

  • Each observation represents a fire detection point recorded on a satellite for a specific time of a date at one location.

  • The variables provided are:

    • Longitude
    • Latitude
    • Observation date (year & Julian day 0-365)
    • Observation time (in UTC)
    • Satellite/sensor used
    • Method of detection (either manual or automated)
    • Ecosystem type (explained below)
    • Fire Radiative Power (FRP) in megawatts

  • In particular, FRP is the rate of emitted radiative energy by the fire at the time of the observation, expressed in units of power (i.e. Watts). It can be used to gauge fire intensity.

  • There is correspondence between method of detection and satellites:

    • VIIRs: SUOMI NPP, NOAA20
    • FDC: GOES-EAST
    • MODIS: MODIS-TERRA

Ecosys

  • The different types of ecosystems are listed in the GLCC README file under Appendix 1 “Global Ecosystems Legend”. These categories have been created based on ” 1-km AVHRR (Advanced Very High Resolution Radiometer) 10-day NDVI (Normalized Difference Vegetation Index ) composites”. The ones we have in the dataset (during the Californian fire) are:

  • From the bar chart, we can see that the following ecosystems are the most prominent during the Californian fire:

    • 22: Cool Conifer Forest
    • 24: Mixed Forest
    • 26: Deciduous Broadleaf Forest
    • 46: Mediterranean Scrub
    • 91: Woody Savanna

Smoke

  • Refer to the dataset smoke to see details.

  • Each observation represents a unique smoke polygon of a specific level recorded at a certain time by one satellite.

  • The variables included in the smoke dataset are:

    • Density: thickness of smoke
    • Satellite/sensor used
    • Geometry: smoke polygon object
    • Type: level of smoke (light, medium, heavy)
    • Start time
    • End time
    • Area: area of the smoke polygon in \(m^2\)

Clustering

Cluster Summary for 11/08/2018-11/16/2018
Cluster Number of Observations Cluster Area Cluster Density Average FRP FRP Variance
2018-11-08_1 13 1.423410e+07 [m^2] 0 [1/m^2] 2.0753 0.5137
2018-11-08_10 10 3.176393e+08 [m^2] 0 [1/m^2] 4.2535 2.6519
2018-11-08_11 7 5.394937e+07 [m^2] 0 [1/m^2] 1.1526 0.2596
2018-11-08_12 6 8.113654e+05 [m^2] 0 [1/m^2] 9.1422 61.9808
2018-11-08_13 7 3.657163e+06 [m^2] 0 [1/m^2] 7.3448 62.7275
2018-11-08_14 9 9.207893e+05 [m^2] 0 [1/m^2] 24.1770 403.1557
2018-11-08_15 31 1.501041e+08 [m^2] 0 [1/m^2] 14.9712 106.6972
2018-11-08_16 8 1.365842e+08 [m^2] 0 [1/m^2] 2.4090 0.5152
2018-11-08_17 15 1.850766e+08 [m^2] 0 [1/m^2] 2.3319 2.0031
2018-11-08_18 2286 2.511210e+08 [m^2] 0 [1/m^2] 134.3567 85643.1991
2018-11-08_2 17 5.667023e+05 [m^2] 0 [1/m^2] 31.8651 1984.3889
2018-11-08_3 20 1.167228e+05 [m^2] 0 [1/m^2] NaN NA
2018-11-08_4 7 3.398569e+06 [m^2] 0 [1/m^2] NaN NA
2018-11-08_5 50 4.815113e+06 [m^2] 0 [1/m^2] NaN NA
2018-11-08_6 11 3.591449e+05 [m^2] 0 [1/m^2] 3.7036 9.9555
2018-11-08_7 38 1.202311e+06 [m^2] 0 [1/m^2] 16.6623 312.9396
2018-11-08_8 38 1.925391e+06 [m^2] 0 [1/m^2] 20.3447 884.1141
2018-11-08_9 9 2.235926e+06 [m^2] 0 [1/m^2] 1.8465 1.1100
2018-11-09_1 10 2.049449e+06 [m^2] 0 [1/m^2] NaN NA
2018-11-09_2 1599 1.002942e+08 [m^2] 0 [1/m^2] 58.6863 13224.1577
2018-11-09_3 2074 1.776215e+08 [m^2] 0 [1/m^2] 104.9371 54726.3656
2018-11-09_4 97 1.406530e+07 [m^2] 0 [1/m^2] 5.8576 38.0230
2018-11-09_5 169 2.571885e+08 [m^2] 0 [1/m^2] 6.9627 49.2537
2018-11-09_6 4018 2.732335e+10 [m^2] 0 [1/m^2] 32.2106 5520.4299
2018-11-10_1 105 3.081300e+06 [m^2] 0 [1/m^2] 18.7275 1102.8667
2018-11-10_10 1603 3.844738e+08 [m^2] 0 [1/m^2] 20.9242 1750.0464
2018-11-10_2 95 5.500144e+06 [m^2] 0 [1/m^2] 3.3426 10.0919
2018-11-10_3 13 5.084765e+05 [m^2] 0 [1/m^2] 3.0239 6.8927
2018-11-10_4 99 3.800582e+06 [m^2] 0 [1/m^2] 7.5454 24.7879
2018-11-10_5 93 1.838807e+06 [m^2] 0 [1/m^2] 20.6890 334.9789
2018-11-10_6 88 3.369599e+06 [m^2] 0 [1/m^2] 15.3668 149.7147
2018-11-10_7 604 2.699461e+07 [m^2] 0 [1/m^2] 14.3358 407.3974
2018-11-10_8 166 9.642361e+06 [m^2] 0 [1/m^2] 4.7362 13.0285
2018-11-10_9 14 4.280154e+05 [m^2] 0 [1/m^2] 4.5693 11.1143
2018-11-11_1 1967 8.197540e+08 [m^2] 0 [1/m^2] 27.2822 8698.4072
2018-11-11_2 274 1.957522e+07 [m^2] 0 [1/m^2] 85.8931 71447.4052
2018-11-11_3 128 9.327657e+06 [m^2] 0 [1/m^2] 38.6417 1660.4962
2018-11-11_4 114 1.112865e+07 [m^2] 0 [1/m^2] 5.4604 24.3280
2018-11-11_5 14 6.121560e+05 [m^2] 0 [1/m^2] 3.9945 22.6090
2018-11-11_6 12 5.606970e+07 [m^2] 0 [1/m^2] 1.9680 2.8212
2018-11-12_1 80 1.806843e+07 [m^2] 0 [1/m^2] 5.0002 18.5542
2018-11-12_2 62 8.770464e+06 [m^2] 0 [1/m^2] 9.7789 97.4233
2018-11-12_3 113 8.374984e+06 [m^2] 0 [1/m^2] 3.7706 18.7874
2018-11-12_4 50 2.184732e+06 [m^2] 0 [1/m^2] 9.8772 191.4502
2018-11-12_5 2683 1.519259e+09 [m^2] 0 [1/m^2] 15.7495 600.7126
2018-11-13_1 99 2.005938e+09 [m^2] 0 [1/m^2] 8.5016 33.9493
2018-11-13_2 44 5.901150e+06 [m^2] 0 [1/m^2] 2.9030 4.4938
2018-11-13_3 34 1.348693e+06 [m^2] 0 [1/m^2] 4.8680 11.0120
2018-11-13_4 546 1.096001e+09 [m^2] 0 [1/m^2] 5.6562 23.5390
2018-11-14_1 128 1.235797e+06 [m^2] 0 [1/m^2] 13.3365 206.0780
2018-11-14_10 56 8.396809e+06 [m^2] 0 [1/m^2] 2.8089 6.2422
2018-11-14_2 174 1.100831e+07 [m^2] 0 [1/m^2] 8.8678 167.3056
2018-11-14_3 364 1.426946e+07 [m^2] 0 [1/m^2] 13.1817 205.8078
2018-11-14_4 128 6.347390e+06 [m^2] 0 [1/m^2] 3.5365 9.0063
2018-11-14_5 19 1.138324e+06 [m^2] 0 [1/m^2] 3.2985 2.5204
2018-11-14_6 14 3.621616e+05 [m^2] 0 [1/m^2] 5.1807 69.5815
2018-11-14_7 26 1.196365e+06 [m^2] 0 [1/m^2] 5.7311 32.5206
2018-11-14_8 56 2.809779e+08 [m^2] 0 [1/m^2] 17.1428 457.6125
2018-11-14_9 22 1.284046e+06 [m^2] 0 [1/m^2] 2.8535 4.9944
2018-11-15_1 270 1.769762e+08 [m^2] 0 [1/m^2] 7.5457 118.7346
2018-11-15_2 32 1.990718e+06 [m^2] 0 [1/m^2] 2.3298 1.9905
2018-11-15_3 196 2.581558e+07 [m^2] 0 [1/m^2] 13.2800 331.1545
2018-11-16_1 19 2.129944e+06 [m^2] 0 [1/m^2] 1.5167 0.7461
2018-11-16_2 62 3.057250e+06 [m^2] 0 [1/m^2] 13.6501 566.9059
2018-11-16_3 34 2.637650e+06 [m^2] 0 [1/m^2] 5.4419 50.2059
2018-11-16_4 14 3.714076e+05 [m^2] 0 [1/m^2] 6.0042 28.5806
2018-11-16_5 23 9.402680e+05 [m^2] 0 [1/m^2] 3.2401 8.6760
2018-11-16_6 276 9.934960e+06 [m^2] 0 [1/m^2] 9.5678 174.1921
2018-11-16_7 130 5.384844e+06 [m^2] 0 [1/m^2] 8.2047 60.4990
2018-11-16_8 173 3.733934e+06 [m^2] 0 [1/m^2] 8.7509 71.7737

Merging with Air Quality System (AQS) data

Smoke Indicator Variables & Distance to Closest Fire

  • For each PM2.5 observation, we will create three indicator variables based on whether it is (1) or is not (0) in the 1) light, 2) medium, and/or 3) heavy smoke plume of the same day.

  • We will also create a numerical variable indicating the distance to the closest cluster (as indicated by a single representative point) in meters.

  • Here’s the first five observations of the augmented AQS dataset (variable aqs):

AQS Dataset for 11/08/2018-11/16/2018
PM25 POC Date Latitude Longitude State County City State.Code County.Code Site.Num Site.Code in_light in_med in_heavy fire_dist closest_cl geometry
14.54167 3 2018-11-08 37.68753 -121.7842 California Alameda Livermore 6 1 7 06-001-0007 1 1 0 60855.64 [m] 2018-11-08_15 POINT (-121.7842 37.68753)
74.20833 3 2018-11-09 37.68753 -121.7842 California Alameda Livermore 6 1 7 06-001-0007 1 1 1 237850.58 [m] 2018-11-09_6 POINT (-121.7842 37.68753)
72.29167 3 2018-11-10 37.68753 -121.7842 California Alameda Livermore 6 1 7 06-001-0007 1 1 1 225617.57 [m] 2018-11-10_7 POINT (-121.7842 37.68753)
61.16667 3 2018-11-11 37.68753 -121.7842 California Alameda Livermore 6 1 7 06-001-0007 1 1 0 240824.38 [m] 2018-11-11_1 POINT (-121.7842 37.68753)
77.79167 3 2018-11-12 37.68753 -121.7842 California Alameda Livermore 6 1 7 06-001-0007 1 0 0 219628.85 [m] 2018-11-12_5 POINT (-121.7842 37.68753)
50.58333 3 2018-11-13 37.68753 -121.7842 California Alameda Livermore 6 1 7 06-001-0007 0 0 0 240868.56 [m] 2018-11-13_4 POINT (-121.7842 37.68753)

Visualization

1. Fire Clusters

  • Here, we take a look at a sample of how polygons are created for each fire cluster. These are the clusters for 11/08/2018.

2. Looking at AQS, fire representatives, and smoke polygons altogether

  • In this visualization, we have smoke polygons from all days 11/8-11/16, fire representative points as circles (colored by day), and AQS PM2.5 observations marked by the grey icons. From this, we can visually see how PM2.5 values reflect their distance to a nearby fire cluster as well as placement in smoke polygons.

```